Infographic: How Accurate Are AI Chatbots?
This chart shows the share of answers from the following AI assistants (free versions) that contain significant/minor inacurracy (in Dec. '24 vs. May-Jun. '25)
Deep Research: One of ChatGPT’s Best Features
Automate web research and get comprehensive reports in 5–30 minutes
AI-text-detectors can be evaded using simple tricks in academic writing, as repeatedly shown by folks like Dr Mike Perkins and Dr Mark A.
AI-text-detectors can be evaded using simple tricks in academic writing, as repeatedly shown by folks like Dr Mike Perkins and Dr Mark A. Bassett. Advice for how to do this is abundant on YouTube, in videos aimed at students, viewed millions of times. Some videos are about how to ‘cheat’, but others have more positive titles like ‘how to study with AI’.
Is there any point trying to stop students using AI to write essays? Or even any value to using asynchronous written essays as summative assessments?
New paper from the great Tomas Foltynek and some bloke called Phil Newton
https://rdcu.be/eKCko
University students feel ‘anxious, confused and distrustful’ about AI in the classroom and among their peers
Whether students and faculty are actively using AI or not, it is having significant interpersonal, emotional effects on learning and trust in the classroom.
An explanation of the “illusion of thinking” paper re: LLMs.
MYFest2024: Skeptical Approaches to AI Research Tools
Anna Mills facilitates the MYFest session held on 2 July 2024. This interactive workshop surveys a range of AI-enabled research assistance apps that aim to help us find and analyze sources. We’ll look at general-purpose ones like Perplexity, ChatGPT4o, and Gemini as well as apps geared to academic research such as Elicit, Consensus, Keenious, ResearchRabbit, SciSpace, Scite_, and Undermind. In what ways do they facilitate source retrieval and analysis, and how can they also mislead us? What does wise use of these tools look like?
Session slides here: https://bit.ly/SkepticalAIresearch